Performance-informed learning effectiveness prediction for customized higher education: an engineering perspective

Abstract

Artificial intelligence (AI) offers a powerful approach to analyze teaching and learning data, which has brought the promising field of AI in education (AIEd) and, particularly, opened new opportunities, potentials and challenges for higher education. The capability of AI has made AIEd widely sought-after as an accurate and efficient approach to predict learning performance while designing teaching strategy. However, it remains challengeable to obtain customized higher education to fit specific requirements of students and satisfy their personalized needs. Recent development in AI algorithms has made it possible to realize automatic update of students' learning performance information, i.e., performance-informed AI (PI-AI). Here, this study first overviews the debut and recent development of AI in higher education, highlighting the PI-AI strategy to maximize teaching and learning performance. Next, we specifically develop a learning effectiveness-informed genetic programming (LEI-GP) model to showcase the application of PI-AI in a case study of a higher ocean engineering course, building on the experimental results of our study, which demonstrated that the LEI-GP model's accuracy in predicting student performance is reasonable, with a maximum Mean Absolute Error (MAE) of 5%. The emerging LEI-GP model is updated with the physiological data of students in learning, which is connected to a high-performance chip system to address the learning data in a real-time wireless manner. Eventually, we provide insights into the PI-AI in propelling real-life customized higher ocean engineering education. PI-AI is an emerging scientific direction in AIEd, which is expected to balance the dilemma between the generalized and customized learning in higher engineering education and address the concern on the design and optimization of its instructional design and teaching strategy.